All-in EV

[shades211]

I apologize in advance for the length of this post. It’s been about 2 years in the making.

I am a long time poker player. I don’t claim to be a great poker player, but I hold my own and have shown a profit over the long term, both live and online. I’m not a professional player, but a solid rec player who plays to make some extra cash. I play in the order of 1000 hands a day (RUSH primarily), not extreme by online standards, but enough that over the course of a year results are telling. I understand the volatility of the game and don’t complain about short term downswings or any particular bad beats.

I have however noticed a long term trend in my results playing online. My actual results consistently trend below my theoretical all-in EV. I noticed the trend well over a year ago. It was so consistent that I assumed PT must be calculating something wrong, so I put together a spreadsheet to calculate it out myself. For the past year I’ve entered every all-in hand I’ve played into this spreadsheet. Yes, over 2000 hands over the course of the year. The numbers matched up pretty much exactly with PT’s numbers.

Long term you should be at your all-in EV. Whether you’re getting your money in good or bad, once the money goes in you have some equity in the pot and the flip of the remaining cards decides how much of that equity you realize in actual profit or loss. Over the long term your actual results should reflect your all-in EV. If you are above your all-in EV you’ve been getting a good run of cards, if you are below your all-in EV you’ve been getting a bad run of cards.

For the past year I'm up 2674BB, with an all-in EV of 5857BB, leaving me 3183BB below my all-in EV. These results are similar to my results going back 2 years (I just don’t have the exact numbers for the previous year).

Over the year I’m hovering around a win rate of just under 1BB/100 hands played. Not great by any means, but positive, then add in rakeback and bonuses and it’s a worthwhile hobby. During that same time my all-in EV line is a little over 2BB/100.

Looking at it another way:

• All-in as a big favorite (>80% equity): 702 wins, 67 loses, win rate of 91%
• All-in as a moderate favorite (between 60 and 80% equity): 262 wins, 94 loses, win rate of 74%
• All-in in a flip situation (between 40 and 60% equity): 141 wins, 181 loses, win rate of 44%
• All-in as a moderate dog (between 20 and 40% equity): 52 wins, 272 loses, win rate of 16%
• All-in as a big dog (between 0 and 20% equity): 25 wins, 464 loses, win rate of 5%

If I get my money in good, my win rate seems reasonable. If I get my money in bad, its pretty much gone. And flips just seem to lean the wrong way. Over the long run this is obviously tough on the bottom line and psychology.

I have one other buddy who plays (or played) a fair bit online. He is also a long term winning player at the tables. He noticed the same effect to the point he gave up playing online.

I am curious if other long term winning players have noticed the same effect? Some poor sucker has to be to the far left of the standard deviation curve, maybe that’s just me for the past couple years. But if this is something that others are seeing as well I think maybe its time to focus on the live games. I’ve been living on blind faith that the games are legit, but seriously, how long can a guy play by simply hoping that eventually the deck won’t be stacked against you.

[travz21]

There's a guy with your exact skill level and sample size that will be that much above all-in EV. He steals it from you.

That's also what happens when people ride it up to the high stakes in a short amount of time. They are good players, but they will play above their bankroll and steal peoples' all-in EV and just crush every level they play in a small sample size. Then they hit high stakes with a decent size roll. Some players lose that and some stay there. Isildur is probably the former.

[grimReaper]

Allin EV does not have to come close to realized winnings, this is misconception. The difference between allin-EV and realized is a random walk with expectation 0 (b/c when you don't go all in, there 0 chance in the difference, and when you go all in, the expected change is 0).

However, your realized winrate and allinEV winrate should come together and converge to your true winrate, as they are both unbiased estimators of the true winrate.

[RAW_FORCE]

64 buyins under is pretty bad but tons of players have experienced worse. I ran 80-90 BI under during my worst stretch but I am on the comeback trail. I know my playing level took a hit when I was running so poorly but it's very important for you to try your best and still play your A game. Obviously your true winrate is made up of how well you play when running normally and when you cant win a pot for months on end.

[Squawk]

Something quite revealing here. If we look at your figures the issue is, as you say, flips and getting the money in bad. You have approx 1000 hands in those situations. To bring the numbers up, you need less than 100 of those hands to have gone the other way (from loss to win) for the numbers to be about right.

When you look at it that way it's pretty easy to see that variance of your all in EV is going to be high even with a hand sample in the hundreds of thousands. Quite an insight that, just opened my eyes up a bit. Thanks